Human milk is a complete source of nutrition for infants as in addition to nutrients it is capable to protect the new-born against several infections in the first months of life providing growth factors and immunological components, which impact on the new-borns’ health(Reference Novak, Almeida and Vieira1). Colostrum is the first fluid secreted by the maternal mammary gland after delivery, and it is the breastmilk most enriched in protective factors such as secretory immunoglobulin A, lactoferrin, leukocytes, bioactive peptides and growth factors. Transitional milk represents mammary gland secretion transitioning between colostrum and mature milk, and it preserves some of the colostrum compositional features; however, it reflects a period of accelerated milk production to meet the nutritional and developmental needs of the growing new-born. Compared with mature milk, colostrum has higher concentrations of proteins, minerals and vitamins, as well as protective factors such as antibodies, immune cells and bioactive peptides providing for the new-born baby and lower levels of lactose, fat and B vitamins(Reference Novak, Almeida and Vieira1,Reference Lönnerdal2) .
Milk proteins are classified into three groups: caseins, whey proteins and mucins. Mucins are the lowest concentrated proteins in milk, and with contents varying from 1 % to 4 % of total proteins during lactation they are found mostly in the membranes of milk fat globules. Whey proteins make up approximately 60 % to 90 % of the total human milk proteins. Regarding caseins, the predominant isoforms in human milk are beta-casein and kappa-casein, which consist of approximately 50 % and from 20 % to 27 % of caseins in breast milk, respectively(Reference Lönnerdal2,Reference Lönnerdal3) . All the proteins’ families in breast milk will potentially give rise to biologically active peptides, which have been shown to present antimicrobial, opioid, immunomodulatory, antihypertensive, antioxidant, antithrombotic, angiotensin-converting enzyme (ACE) and dipeptidyl peptidase IV inhibitory activities that influence the infants’ immune status and development(Reference Lönnerdal3–Reference Wada and Lönnerdal5).
Breast milk imparts antimicrobial peptides (AMP) to the new-born, providing defence against pathogens. Human milk protein-derived peptides can be released via three major pathways: (i) proteolysis by milk endogenous proteases, such as plasmin; (ii) proteolysis by digestive enzymes, such as pepsin, trypsin and chymotrypsin and (iii) proteolysis by proteases derived from proteolytic microorganisms in the breast milk microbiome(Reference Korhonen and Pihlanto6). AMP are a class of multifunctional defence molecules, and their role in breast milk as a defence line protecting the new-born against infection remains largely uninvestigated(Reference Lönnerdal2,Reference Khan, Pirzadeh and Förster7) . In fact, there is much to be known about the defence mechanism provided to the new-born through milk. For instance, The WHO encourages breast-feeding even when the lactating mother has a viral infection, such as HIV infection or COVID-19(Reference Gribble, Mathisen and Ververs8), as the benefits of breast-feeding overcome the risks of infection.
Bioactive peptides released during in vitro digestion of mature breast milk have been previously investigated(Reference Dallas, Murray and Gan9–Reference Wada and Lönnerdal11). To the best of our knowledge, the present study is the first one to reveal the release of bioactive peptides from proteins in human colostrum during in vitro gastrointestinal digestion annotating potentially bioactive peptides that were released, with emphasis on AMP that may impact the new-borns’ ability to protect against microbes, and indicating which proteolytic enzymes acted to lead to the formation of these peptides.
Material and methods
Subjects and sample collection
Mothers (21–40 years) who delivered at term were recruited and colostrum milk samples were obtained from twenty-four volunteers. Milk samples (on average 3 ml; 1–5 ml) were collected by manual expression into polypropylene containers in the Herculano Pinheiro Maternity Hospital in Rio de Janeiro. Samples were collected between November 2017 and January 2018. Participants were recruited a few weeks before birth, delivered full-term infants (37 to 41 weeks) and manually expressed colostrum samples between 1 and 5 d postpartum. All participants were eutrophic at the start of pregnancy. Milk collection followed standardised procedures, was performed in the morning before breast-feeding (foremilk) and the volunteers’ breasts and hands were thoroughly washed with soap and rinsed with deionised water prior to each sampling (Fig. 1). Exclusion criteria were (1) chronic health problems, (2) pregnancy complications, (3) smokers and alcohol drinkers during pregnancy and (4) users of nutritional supplements (except Fe and folate). Milk samples were transported on ice to the laboratory and stored at −80°C until analysis. Before storage, samples were randomly selected for microbiological analysis to confirm that the collection protocol adopted was appropriate for sample preservation aiming at peptidomics analysis. All volunteers in this study were from a similar socio-economic background regarding educational level (number of years of formal education) and monthly household income (varying from less than one to up to three Brazilian minimum wages; during recruiting a minimum wage in Brazil corresponded to USD 300 per month), assessed by standardised questionnaires applied during recruiting.
Fig. 1. Schematic representation of the study workflow. Human colostrum was digested in vitro to investigate the release of antimicrobial peptides. LC-HRMS /MS analysis of peptide samples were performed using a Q-Exactive instrument (Orbitrap-based MS). Promising candidates of antimicrobial peptides were prospected via bioinformatic approaches. LC-HRMS, liquid chromatography-high resolution MS
This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Ethical Committee of Pedro Ernesto University Hospital, State University of Rio de Janeiro (Protocol number CAAE: 56617516·5·0000·5259). Written informed consent was obtained from all subjects.
Sample preparation
Colostrum samples were thawed in a water bath at 37°C, and protein concentration was determined according to the Lowry(Reference Lowry12) protein assay. The colostrum samples from the twenty-four donors were pooled; the protein composition in the pool was balanced and contained the same amount of protein (30·0 mg) from the milk of each donor. A total volume of 16 ml of pooled sample was used for the in vitro digestion experiments.
In vitro digestion
Simulated gastrointestinal digestion experiments were conducted as described by Versantvoort(Reference Versantvoort, Oomen and Van de Kamp13) and adapted from Dall’Asta(Reference Dall’Asta, Florio and Lammardo14) that uses pH values typically seen in new-borns, running sequential digestive phases in three parallel 50 ml centrifuge tubes (triplicate). The artificial digestive fluids were formulated (online Supplementary Table 1) and were equilibrated at 37 ± 2°C in a water bath before starting the simulated digestions, which were all run at 230 rpm at 37 ± 1°C under constant shaking/stirring. The digestion started by adding 1 ml of a salivary solution containing α-amylase to 5·0 ml of pooled human colostrum and incubating for 5 min. Then, the gastric phase was started by adding 2 ml of the gastric fluid containing pepsin and adjusting the pH value to 4·5 with 1 mol/l HCl; samples were then incubated for 60 min. Finally, the intestinal phase started by sequentially adding 2 ml of duodenal fluid, 1 ml of bile salts, 0·3 ml of bicarbonate (1 mol/l NaHCO3) and 1 ml of β-galactosidase solutions; after adjusting the pH to 7·5, samples were incubated for additional 60 min. For each phase and considering the continuous-time recording, milk digest from 5 min (saliva), 35 and 65 min (mid and end of the gastric phase) and 125 min (duodenal phase) were sampled in duplicate from each tube and centrifuged (14 000 × g, 4°C, 20 min) using Microcon concentrators (YM-10K, Millipore), and the permeates containing the peptides were collected for analysis by mass spectrometry. Peptides’ content in each phase was determined by the Qubit™ Quantitation Fluorometer (Invitrogen; Thermo Fisher Scientific, Inc.).
Mass spectrometry analysis and data processing
The samples were injected in a nano-LC Ultimate 3000 (Thermo Fisher Scientific) coupled to a Quadrupole-Orbitrap Q-Exactive Plus mass spectrometer (Thermo Fisher Scientific, Bremen, GER). A total of three samples from each digestion phase were analysed as triplicates from the in vitro digestion. One microgram of peptides was loaded on a pre-column (2 cm length, 200 µm inner diameter), packed with ReproSil-Pur C18-AQ 5 µm resin and fractionated in an analytical column Picochip with ReproSil-Pur C18 3 µm resin. Peptides were eluted using a gradient of 95 % solvent A (95 % H2O, 5 % ACN, 0·1 % formic acid) to 40 % solvent B (95 % ACN, 5 % H2O, 0·1 % formic acid) for 60 min; 40 % to 85 % of solvent B for 10 min and 95 % of solvent B for 15 min, with a flow at 300 nl/min. Data were acquired in full scan and MS/MS acquisition in positive mode (ESI+), and the MS/MS data were obtained by data-dependent acquisition (dd-MS2). The ion source and S-lens were optimised to spray voltage at 3·4 kV, zero flow of sheath and auxiliary gas at 250°C and 80 S-Lens RF level. MS full scan was acquired at 70 000 Hz resolution between m/z 400 and 2000 in the Orbitrap analyser, 106 automatic gain control and 50 ms maximum ion injection time. For the fragmentation experiments, a DDA protocol was adopted in which the ten most intense ions with charges higher than 2 were selected for high-energy collision dissociation fragmentation. Fragment acquisition in the Orbitrap using thirty collision energy and dynamic exclusion was enabled for 20 s. The MS/MS scans were obtained using 17 500 Hz resolution, 5 × 104 AGC, 50 ms maximum isolation time, 1 microscan/sec, scanning between m/z 200 and 2000, in an m/z 2·0 isolation window. The DDA raw data were processed and searched by the Proteome Discovery 2·1 software server search engine (Thermo Fisher Scientific, Bremen, GER) with the SEQUEST algorithm by using a tolerance up to ± 0·1 Da to precursor ions and 0·01 Da to fragment ions. Protein identification was performed by searching the mass spectrometric data against the UniProt-SwissProt protein database (available in April 2018) containing reversed sequences with a false discovery rate < 1%. Small peptides, such as those with two to four amino acid residues, remained undetected or unprocessed and this was important to avoid reporting of misannotated peptides.
Physicochemical and enzymatic hydrolysis analysis
The ProtParam Tool (http://web.expasy.org/protparam) was used to compute the amino acid composition (%) and to estimate the peptides’ pI and theoretical molecular weight, and the hydrophobicity was estimated using Gravy Calculator (http://www.gravy-calculator.de/). The proteolysis pattern was predicted by Enzyme-Predictor(Reference Vijayakumar, Guerrero and Davey15) (http://bioware.ucd.ie/∼enzpred/Enzpred.php), and peptide release was visualised by the software CAITITU tool(Reference Carvalho, Junqueira and Valente16) (version 2.0.0.19; available free to download from: http://pcarvalho.com/things/caititu/).
To determine the patterns of proteolytic activity, the peptide sequences obtained after in vitro digestion and annotated by mass spectrometry were uploaded in the EnzymePredictor(Reference Vijayakumar, Guerrero and Davey15) (http://bioware.ucd.ie/∼enzpred/Enzpred.php), together with the input of proteolytic enzymes present during each phase of digestion (endogenous proteases and those from the simulated digestive fluids), and the enzymes’ hydrolytic patterns (http://bioware.ucd.ie/∼enzpred/patterns.html). EnzymePredictor is an online bioinformatics tool that identifies the proteolytic enzymes that most likely acted in the formation of each peptide produced from the hydrolysis of a mixture of proteins and when a combination of proteolytic enzymes was used.
Prospection of antimicrobial peptides
Peptide sequences obtained by mass spectrometry were compared with sequences from three open functional peptide databases: Milk bioactive peptide database(Reference Nielsen, Beverly and Qu17) (http://mbpdb.nws.oregonstate.edu/), AMPA(Reference Torrent, Di Tommaso and Pulido18) (Antimicrobial Sequence Scanning System; http://tcoffee.crg.cat/apps/ampa/do) and ClassAMP(Reference Joseph, Karnik and Nilawe19) (a prediction tool for classification of AMPs; http://www.bicnirrh.res.in/classamp/). These bioinformatics tools use different approaches to predict the bioactivity of peptides. Milk Bioactive Peptide Database is a databank and search tool on functional milk peptides from different mammalian species from published peer-reviewed studies(Reference Nielsen, Beverly and Qu17). AMPA identifies putative antimicrobial regions in the peptide sequences, supporting the prospection and design of antimicrobial peptides(Reference Torrent, Di Tommaso and Pulido18). In contrast, ClassAMP is used to predict the propensity of peptides to have antibacterial, antifungal or antiviral activity(Reference Joseph, Karnik and Nilawe19).
Statistical analysis
Results were reported as the mean ± standard deviation of triplicate determinations. The statistical analysis was carried out using the software GraphPad Prism (version 8.0.1, GraphPad Software). Peptide overlap was analysed using Venn diagram generated by the online tool Venny (https://bioinfogp.cnb.csic.es/tools/venny/).
Data availability
The proteomics datasets produced in this study are available in the PRIDE database (https://www.ebi.ac.UK/pride/) at the following accession code: PXD035066. The data is currently private (publication pending) and will be made available when this work is accepted.
Results
Human milk samples were provided by twenty-four healthy mothers after informed consent. Supplementary Table 2 shows the characteristics of the donors and their new-borns during peri-conception. Protein concentrations in each colostrum sample varied from 1·85 to 13·07 g/100 ml. This interindividual variation was accounted for by combining all samples in a balanced sample pool containing the same amount of protein from each donor milk.
Peptide release
The formation of peptides during in vitro gastrointestinal digestion was monitored by quantitative fluorimetry and high-resolution mass spectrometry and MS/MS analysis (Fig. 1). As expected, a lower concentration of peptides was seen during the oral phase of digestion (Fig. 2(a)) and the duodenal phase of digestion presented the peptides with the lowest number of amino acid residues and consequently the smallest masses (Fig. 2(b)). In total, 2318 peptide sequences were obtained from 112 precursor proteins (online Supplementary Table 3). The Venn diagram (Fig. 2(c)) shows the number of peptides identified per phase of in vitro colostrum digestion and some peptides were unique to a single phase of digestion.
Fig. 2. Peptide profiling in human colostrum. (a) Colostrum peptide concentration at each phase of simulated in vitro digestion was determined fluorometrically. (b) Average mass (Da) and number of amino acid residues of the peptides identified. (c) Human colostrum peptidome during in vitro digestion. Venn diagram showing the number of peptides identified per phase of human colostrum simulated digestion. Three 5·0 ml pools containing equal amounts of protein from each of the twenty-four colostrum samples were used for the in vitro digestion experiments. Error bars in (a) and (b) represent sem of the triplicates of in vitro colostrum digestion.
Formation of peptides during the simulated gastrointestinal digestion
The enzyme hydrolysis pattern was predicted by Enzyme-Predictor and was associated with plasmin, trypsin, chymotrypsin, pepsin, cathepsin D, elastase and glutamyl endopeptidase activities, in that order for the number of peptides formed (Fig. 3). The proteins that served as precursors to most peptides formed were caseins and immunoglobulins (online Supplementary Table 3). Additionally, many peptides were derived from the polymeric immunoglobulin receptor (P01833), tenascin (P24821–4), clusterin (P10909–2) and complement C3 (P01024). The CAITITU tool (version 2·0·0·19) was used to follow the formation of peptides during the simulated digestion. This bioinformatics tool predicted the location of the peptides in the primary structure of the precursor protein that originated during in vitro digestion (Fig. 4). Less than 3 % of the peptides annotated, from all the peptides originating from colostrum digestion, potentially derived from hydrolysis of proteins used to formulate the artificial digestive fluids (online Supplementary Table 3).
Fig. 3. Number of proteins cleaved. Absolute frequencies of cleaved colostrum proteins per predicted digestive enzyme.
Fig. 4. Graphic representations of the location of peptides derived from. (a) Polymeric immunoglobulin receptor; (b) clusterin; (c) isoform 4 of tenascin, containing a signal peptide region (SP), and EGF-like, fibronectin type III and C-terminal domains; (d) C3 complement protein, containing a signal peptide region (SP) and a mature protein (alpha and beta chains). These images were created from MS/MS peptidomics data analysed by CAITITU software (version 2.0.0.19).
Peptides with potential antimicrobial activity
Three tools were used for annotation of antimicrobial peptides. By searching in bioactive peptides’ databases, the peptides sequences from milk digestion, previously reported functions were retrieved. Peptides retrieved from these databases were all derived from beta-casein which is the main casein in human milk (Table 1). From the annotated peptides, two showed potential antimicrobial activity (Table 1). Seven peptide sequences with antimicrobial activity were predicted by using AMPA software (Table 2). These peptides were fragments from isoform 4 tenascin proteins, beta-2-microglobulin, complement C3 and protein S100-A8. The predicted peptides’ mass (Da), hydrophobicity and pI computed from more than one database are presented in Table 2. These peptides were all cationic as they contained from two to four arginine and/or lysine residues, and they had from 33 % to 57 % of hydrophobic residues. The location of each of these cationic peptides in the three-dimensional structures of its respective precursor protein is shown in Fig. 5. Peptides arising from tenascin isoform 4 digestion were in a region containing an α-helix, a random coil and a β-sheet that is stabilised by an interchain disulphide bond. The peptide originating from β-2-microglobulin protein presents a region with an α-helix connecting two β-sheets. C3 complement peptides form a two β-sheet structure connected by a random coil. Bioactivity prediction software ClassAMP predicted numerous sequences with antifungal activity (50 % frequency) that were sensibly more frequent than the antiviral (29 %) and antibacterial (21 %) peptides predicted from the 2318 sequences annotated from the high-resolution mass spectrometry data (online Supplementary Table 5).
Table 1. Peptide sequences annotated in in vitro digested human colostrum (n 24), having beta-casein (Uniprot ID P05814) as precursor protein as retrieved from the Milk Bioactive Peptide Database (http://mbpdb.nws.oregonstate.edu/)
* Grey colour cell filling indicates the detection of each peptide; blank cells indicate absence of detectable amounts of the peptides. ACE = angiotensin-converting enzyme.
Table 2. Predicted antimicrobial peptides from the in vitro simulated neonatal digestion of human colostrum (n 24) by using the AMPA bioinformatic tool (tcoffee.crg.cat/apps/ampa): experimental exact mass, isoelectric point (pI) and hydrophobicity, according to the phase of digestion in which they were observed
* Numbers between parentheses indicate the initial and final position of the peptide in the primary sequence of the precursor protein, considering the signal peptide. Codes used for identification of amino acid residues’ properties in the peptide sequences: bold font was used for arginine and lysine (R and K, respectively), which have positively charged side chains; cyano font was used for amino acids with hydrophobic side chains; yellow font was used for cysteine residues, indicating potential sites of disulphide bonds.
Fig. 5. 3D Structure of proteins. (a) Beta-2-microglobulin (2YXF), (b) complement C3 (2A73), (c) isoform 4 of tenascin (6QNV), and (d) protein S100-A8 (1MR8). Colour ribbons represent the locations of the potentially antimicrobial peptides predicted in this study. Figures were visualised with the RASMOL software and the structures obtained from the Protein Data Bank (rcsb.org).
Discussion
The protective role of colostrum against infections is widely reported and helps to prevent life-threatening infections that remain major causes of neonatal death. Protective factors in colostrum include immunoglobulins and bioactive peptides that probably reduce neonatal mortality by infectious diseases. These factors can also actively influence the maturation of the neonatal immune system(20). Peptidomics of mature breast milk revealed the formation of bioactive peptides during proteolytic digestion of milk proteins in previous reports(Reference Armaforte, Curran and Huppertz21,Reference Dallas, Guerrero and Khaldi22) . But to the best of our knowledge, assessing the profile of bioactive peptides formed during the digestion of human colostrum has not been previously reported, and this could be relevant as the new-born infant is most vulnerable to infections.
In this study, we aimed at prospecting bioactive peptides formed during simulated in vitro digestion of human colostrum, to determine how these molecules are formed over time and phase of digestion, either oral, gastric or duodenal. Human colostrum was digested in vitro with porcine pepsin and pancreatin, and the time course of peptides’ formation was assessed via the MS-based peptidomics profile throughout incubation up until 125 min. A dynamic in vitro digestion protocol was used in this study, in which the composition of the digestive fluids was changed from one digestion phase to the other. By this approach, the dynamic changes taking place during digestion could be more closely simulated, but it limits data comparisons with other studies using the official INFOGEST method(Reference Brodkorb, Egger and Alminger23). The peptides were analysed by high-resolution mass spectrometry in a nanoLC-Q-Exactive Orbitrap® platform, leading to the detection of 2318 peptides, originating from 112 precursor proteins. Most of these peptides were observed in the gastric phase, and the smallest peptides were detected in the final phase of digestion. However, a considerable concentration and count of peptides were observed in the oral phase. Only one peptide sequence annotated in the oral phase was also present in colostrum before digestion (endogenous peptide; online Supplementary Table 5). Therefore, in the present study we did not confirm that high contents and/or variety of peptides derived from beta-casein were present in milk before digestion(Reference Ferranti, Traisci and Picariello24–Reference Khaldi, Vijayakumar and Dallas26). It could not be ruled out that the absence of endogenous peptides originating from beta-casein in the present work was related to the analytical strategy adopted to avoid peptides’ misannotation in which single-charged peptides were not detected and fragmented. It is debatable if endogenous peptides in milk are synthesised as such in mammary epithelial cells or if they are formed by the activity of endogenous proteases in the interval between milk secretion and ejection(Reference Guerrero, Dallas and Contreras25,Reference Khaldi, Vijayakumar and Dallas26) . Our results show evidence in favour of the second hypothesis, however, as this hypothesis was not directly tested in this study, it would be of interest for future investigations. Several of the peptides seen in the oral phase remained intact in the end of digestion, such as the sequences LLLNPTHQIYPVTQPLAPV and YPVTQPLAPVHNPIS that were previously shown to be antibacterial(Reference Minervini, Algaron and Rizzello27,Reference Wang, Sun and Wang28) . In contrast, the antimicrobial sequence LLNQELLLNPTHQIYPV was detected solely in the oral phase, indicating that it could have been formed very early in colostrum digestion, possibly by the proteolytic activity of milk proteases(Reference Fu, Ji and Chen29). Amino acid (AA) sequences of these 112 proteins were calculated using the ProtParam tool. A high frequency of the amino acids proline (P), valine (V), leucine (L) and glutamine (Q) was seen in peptides’ sequences varying between thirty and forty-five residues. The abundance of proline residues in a peptide sequence can increase resistance to digestive enzymes(Reference Yang and Russell30) and, consequently, peptides’ residence time in the digestive system.
Previous works have demonstrated that human milk contains proteolytic enzymes such as plasmin, trypsin, elastase, cathepsin D, thrombin, kallikrein and several carboxy- and amino-peptidases(Reference Dallas, Murray and Gan9). These proteases are active within the mammary glands and originate specific peptides via proteolysis, possibly before milk ejection. But little is known about the digestive process in new-borns and how peptides are derived. To predict the enzymatic cleavage patterns, the Enzymer Predictor software was used, and the enzymes’ catalytic activity is shown in online Supplementary Table 4. Proteolysis by plasmin and trypsin leads to the formation of the most numerous protein fragments among the proteolytic enzymes herein investigated (Fig. 3). Plasmin is the most active endogenous protease in human milk that is an endopeptidase cleaving peptide bonds of lysine or arginine residues(Reference Christensen, Schack and Kläning31), and its digestive activity is higher in premature than in term breastmilk(Reference Armaforte, Curran and Huppertz21,Reference Demers-Mathieu, Qu and Underwood32) . The importance of pepsin in the proteolysis of human colostrum was clear from our in vitro experiments, being the protease that led to the formation of the highest number of protein fragments, just slightly higher than trypsin. Trypsin is also present in human milk, and it cleaves peptide bonds of lysine or arginine residues(Reference Dallas, Murray and Gan9). The hydrolytic activity of plasmin and trypsin produced, respectively, 892 and 890 peptides (online Supplementary Table 4), which were derived from 188 colostrum proteins. However, hydrolysis of human colostrum by Cathepsin D cleaved milk proteins in 1030 fragments (online Supplementary Table 4), although it has hydrolysed a lower number of colostrum proteins (130; Fig. 3) compared with plasmin and trypsin. Possibly, the higher number of milk fragments deriving from cathepsin D proteolytic activity is related to its optimum pH value being 5·0 which is near to the pH value (4·5) used in the gastric phase, simulating the acidity expected in the new-borns’ stomach.
The formation of peptides from precursor proteins was graphically represented with the assistance of the CAITITU software (Fig. 4). Although most of the peptides were formed in the gastric phase, this chart (Fig. 4) shows that the duodenal phase produced a greater variety of peptides. In the present work, the digestion behaviour seemed to be at least partly dependent on the protein substrate, as various peptides were derived from the N-terminus of complement C3, while the digestion of tenascin isoform 4 originated peptides from the C-terminus.
The three bioinformatics tools employed in this work provided a thorough characterisation of antimicrobial peptides. The milk bioactive peptide database includes several peptides derived from a variety of milk products and was searched to identify peptides with biological functions. All the peptides investigated in this database (milk bioactive peptide database) were derived from beta-casein, which is the main casein in human milk. There is a persistent need for a database of peptides from human milk, before and after digestion, especially aiming at future research regarding the health of lactating infants. Caseins have been widely studied because of their nutritional role for the new born, as well as for being a precursor of bioactive peptides(Reference Lönnerdal4), formed mostly during proteolysis(Reference Ferranti, Traisci and Picariello24,Reference Silva and Malcata33) . Although two of the bioactive sequences have potential antimicrobial activity, most sequences with predicted bioactivity had ACE inhibitor activity. ACE inhibitors lower blood pressure and modulate inflammation, and milk caseins and whey proteins are known precursors of ACE inhibitory peptides(Reference Wada and Lönnerdal5). A recent study showed that preterm infants having a higher exposure to maternal breastmilk had enhanced cardiac function and morphology at age one year, approaching the values seen in full-term infants at that age(Reference El-Khuffash, Lewandowski and Jain34). The mechanisms leading to these improvements in the cardiac health of infants born preterm that received breastmilk are not fully known. However, it could not be discarded that bioactive peptides in milk or formed during milk protein digestion have a protective role in this sense, especially considering the increased intestinal permeability of preterm infants(Reference Weström, Sureda and Pierzynowska35).
Bioactivity prediction software (ClassAMP) predicted a higher number of antifungal peptides compared with antiviral and antibacterial peptides examined in this study (Fig. 1). These peptides were rich in hydrophobic amino acids and were seen in the gastric phase of digestion at 65 min. A common feature of antimicrobial peptides is the presence of at least 50 % hydrophobic amino acids(Reference Hancock and Scott36). Previous studies have shown that antimicrobial peptides tend to have amphipathic structures and to be positively charged at physiological pH, being these two features also conserved among many antimicrobial peptides(Reference Hancock and Scott36,Reference Yeaman and Yount37) . Physicochemical characteristics of antimicrobial peptides are known to be fundamental to their bioactivity(Reference Tu, Liu and Cheng38).
The regions in the polypeptide chains of human milk proteins with potential antimicrobial activity were calculated by the AMPA bioinformatics tool. Seven sequences were prospected, and the peptides were rich in hydrophobic amino acids. These peptides were fragments of isoform 4 of tenascin, beta-2-microglobulin, complement C3 and protein S100-A8. The structural arrangements of these peptides closely resemble the structure of defensins that are cationic and disulphide-rich antimicrobial peptides with less than 100 amino acids. Furthermore, the amphiphilic peptide from tenascin presents a disulphide bridge that is also a characteristic feature of defensins. The most prominent antimicrobial structures were amphiphilic peptides with two to four β-strands, and amphipathic α-helix and loop structures. Most of the peptides originating from these four proteins (Table 2) have been assigned as having antifungal activity in more than one database. Mechanisms of action of antimicrobial peptides are associated to their structures, so that to a large extent peptide function is determined by its structure(Reference Yeaman and Yount37). Therefore, AMP formed during colostrum digestion have been organised into four functional groups based on the presence or absence of key structural elements such as α-helices, β-sheets and various turns and random coils(Reference Tu, Liu and Cheng38). Taken together, the results from physical and chemical properties of the bioactive peptides derived from the digestive proteolysis of tenascin isoform 4, beta-2-microglobulin, complement C3 and protein S100-A8 (Table 2 and Fig. 4), combined with the depiction of peptides’ 3D structure, and prediction of peptides’ putative antimicrobial activity using bioinformatics tools point to new molecules with potential antimicrobial activity.
Tenascin-C is important to fetal development and wound healing, but antimicrobial properties have not been previously reported. Fouda et al.(Reference Fouda, Jaeger and Amos39) have shown that this protein neutralised HIV-1 by binding to the virus envelope protein at the chemokine coreceptor site of CD4 lymphocytes. The complement C3 protein plays a central role in both the classical and alternative pathways of complement activation. This protein is involved in both inflammatory processes and the antimicrobial response(Reference Koehbach and Craik40,Reference Ogundele41) . Complement C3 is a protein of the human milk complement system that represents an important part of neonatal innate immunity, in addition to having antimicrobial effects. Furthermore, it can act to protect the lactating mammary gland by reducing the risks of infections(Reference Mazumdar, Kim and Meyer42). Protein S36100A8 is a calcium- and zinc-binding protein that exert intra and extracellular functions including roles in Ca2+ homeostasis, cell proliferation and differentiation and proinflammatory activity. Antimicrobial properties of protein S100A8 are possibly related to Zn2+ chelating activity, as this metal is essential for microbial growth(Reference Trégoat, Cuillière and Molé43,Reference Zackular, Chazin and Skaar44) . The immunologic role of beta-2-microglobulin in milk has been previously characterised(Reference Kozlyuk, Monteith and Garcia45,Reference Groves and Greenberg46) . However, S100-A8 regulates inflammatory processes, and the immune system(Reference Werner, Floc’h and Migliore-Samour47), and was detected in milk for the first time in this study, to the best of our knowledge. Digestion of S100-A8 lead to formation of peptide sequence SIIDVYHKYSLIKGNF that has potential antimicrobial activity.
The bioinformatic tools used in this study proved useful to prospect potential bioactive peptides arising from human milk in vitro digestion. As the intestinal permeability in new-borns is higher than in adults until after 1 week after birth at term(Reference Weström, Sureda and Pierzynowska35), bioactive peptides formed during digestion might have systemic effects in vivo if absorbed by the new-born infant. And these potential systemic effects of milk bioactive peptides could be even more pronounced in premature or formula-fed infants that have higher intestinal permeability than breast-fed and term infants(Reference Weström, Sureda and Pierzynowska35). Peptides investigated in the present study have a high potential for biotechnological application, especially for the formulation of improved infant formulae if their bioactivity is confirmed. Future investigations with these potentially bioactive peptides are deserved to confirm bioactivity in vivo, especially the antimicrobial peptides, as drug-resistant pathogens are emerging and spreading, leading to antimicrobial resistance that is increasingly threatening our ability to treat infections. Determining the in vivo bioactivity of the ACE inhibitors would be also of interest, and future research aiming at investigating the effects of milk bioactive peptides on the improvement of premature infants’ cardiovascular health would be relevant. In addition, future studies investigating the formation of bioactive peptides from new precursor proteins in food seem promising in the fields of nutrition and food technology aiming at improved food products for better consumers’ health.
Acknowledgements
The authors thank Dr. Giovani Carlo Veríssimo da Costa for his technical assistance with the Mass spectrometry analysis.
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, CAPES-Brazil (Finance Code 001); Conselho Nacional de Desenvolvimento Científico e Tecnológico, CNPq-Brazil (Grant # 432484/2016-7), Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro, FAPERJ-Brazil (Grants # 203·197/2015, 010·101016/2018 and 010·001436/2019). I.B.C. was a recipient of CAPES/CNPq PhD scholarships, and A.G.T. was a recipient of CNPq and FAPERJ research fellowships.
I. C.: Validation, formal analysis, investigation, writing – original draft, visualisation; P. A.: formal analysis, writing – review and editing; F. B.: writing – review and editing; M. S.: conceptualisation, validation, resources, data curation, writing – review and editing, funding acquisition and A. G. T.: conceptualisation, resources, writing – review and editing, funding acquisition. All authors read and approved the final manuscript.
All authors (I. B. C., P. F. A., F. F. B., M. R. S. and A. G. T.) have no competing interests to declare.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114523001459
